Applied Panel Data Analysis for Economic and Social Surveys

✍ Scribed by Hans-Jürgen Andreß, Katrin Golsch, Alexander W. Schmidt

• Publisher: Springer
• Year: 2013
• Tongue: English
• Leaves: 337
• Category: Library

✦ Synopsis

Many economic and social surveys are designed as panel studies, which provide important data for describing social changes and testing causal relations between social phenomena. This textbook shows how to manage, describe, and model these kinds of data. It presents models for continuous and categorical dependent variables, focusing either on the level of these variables at different points in time or on their change over time. It covers fixed and random effects models, models for change scores and event history models. All statistical methods are explained in an application-centered style using research examples from scholarly journals, which can be replicated by the reader through data provided on the accompanying website. As all models are compared to each other, it provides valuable assistance with choosing the right model in applied research. The textbook is directed at master and doctoral students as well as applied researchers in the social sciences, psychology, business administration and economics. Readers should be familiar with linear regression and have a good understanding of ordinary least squares estimation.

​

✦ Table of Contents

Applied Panel Data Analysis for Economic and Social Surveys
Contents
List of Figures
List of Tables
List of Textboxes
List of Examples
Chapter 1: Introduction
1.1 Benefits and Challenges of the Panel Design
1.1.1 Benefits
1.1.1.1 Measuring Change at the Individual Level
1.1.1.2 Separating Age and Cohort Effects
1.1.1.3 Controlling for Omitted Variable Bias
1.1.1.4 Assessing Causality
1.1.1.5 Obtaining Larger Sample Sizes
1.1.1.6 Measurement Error
1.1.2 Challenges
1.1.2.1 How to Represent the Population over Time?
1.1.2.2 How to Obtain Valid and Reliable Measurements over Time?
1.1.2.3 How Much Does It Cost?
1.2 Outline of the Book
1.3 Audience and Prerequisites
Chapter 2: Managing Panel Data
2.1 The Nature of Panel Data
2.2 The Basics of Panel Data Management
2.2.1 Merging and Appending Data
2.2.2 Basic Append and Merge Commands
2.2.3 Building a Working Data Set with Append and Merge Commands
2.2.4 Wide and Long Format
2.2.5 Some General Remarks
2.3 Three Case Studies on Poverty in Germany
2.3.1 Case Study 1: How Many German Citizens Were Poor in 2004? A Cross-Sectional Analysis
2.3.2 Case Study 2: Did Poverty Increase in Germany After 2004? An Analysis of Pooled Cross-Sections
2.3.3 Case Study 3: How Large Is the Risk of Becoming Poor in Germany? A Panel Analysis
2.4 How to Represent a Population with Panel Data?
2.4.1 Weighted and Unweighted Analysis of Cross-Sections
2.4.2 Weighting in Balanced and Unbalanced Panels
2.4.3 When to Use Weights?
2.5 Conclusion and Further Reading
Chapter 3: Describing and Modeling Panel Data
3.1 Some Basic Terminology
3.2 Measurements over Time Are Not Independent
3.3 Describing the Dependent Variable
3.4 Explaining the Dependent Variable over Time: Typical Explanatory Variables
3.4.1 Time-Constant and Time-Varying Variables
3.4.2 Serially Correlated Observations
3.5 Modeling Panel Data
3.5.1 Modeling the Level of the Dependent Variable
3.5.1.1 Continuous Dependent Variables
3.5.1.2 Categorical Dependent Variables
3.5.2 Modeling Change of the Dependent Variable
3.5.2.1 Continuous Dependent Variables
3.5.2.2 Categorical Dependent Variables
3.5.3 Additional Models
3.5.3.1 Other Types of Relationship
3.5.3.2 How to Include Time-Constant Variables in Models of Change
3.5.3.3 How to Relax the Assumption of Fixed Coefficients
3.6 Estimating Models for Panel Data
3.6.1 Omitted Variable Bias (Unobserved Heterogeneity)
3.6.1.1 What Is the Problem?
3.6.1.2 How to Extend Panel Regression Models for Unobservables
3.6.1.3 Why Are Panel Data Useful to Control for Unobserved Heterogeneity?
3.6.2 Serially Correlated and Heteroscedastic Errors
3.6.3 Measurement Error Bias
3.6.3.1 Measurement Error in the Dependent Variable
3.6.3.2 Measurement Error in the Explanatory Variables
3.6.3.3 Structural Equation Models
3.6.4 A Formal Summary of the Main Estimation Assumptions
3.7 Overview of Subsequent Chapters
Chapter 4: Panel Analysis of Continuous Variables
4.1 Modeling the Level of Y
4.1.1 Ignoring the Panel Structure
4.1.1.1 Pooled Ordinary Least Squares
4.1.1.2 Robust Standard Errors
4.1.2 Modeling the Panel Structure
4.1.2.1 Correlated Heterogeneity: Fixed Effects Estimation
FE Estimation Using Dummy Variables
FE Estimation Using Time-Demeaned Data
When and How Do FE Estimates Deviate from Pooled OLS Estimates?
How Does FE Estimation Control for Serial Dependence?
4.1.2.2 Uncorrelated Heterogeneity: Random Effects Estimation
4.1.2.3 Combining Fixed and Random Effects Estimation: A Hybrid Model
Relationship Between Pooled OLS, FE, and RE
When to Apply Pooled OLS, FE, and RE?
A Hybrid Model
Testing Differences Between RE and FE Estimates
4.1.2.4 Wrapping Up: How to Choose Between the Different Models in Applied Panel Research?
4.1.3 Extensions
4.1.3.1 Models with Random Intercepts and Random Slopes
4.1.3.2 More Complicated Error Processes
4.1.3.3 Unbalanced Panel Data
4.1.3.4 Models for Data in Wide Format
4.2 Modeling the Change of Y
4.2.1 Analysis of Change Using Change Scores
4.2.2 Analysis of Change Using Impact Functions
4.2.3 Analysis of Trends
4.3 Conclusion and Further Reading
Chapter 5: Panel Analysis of Categorical Variables
5.1 Modeling the Level of Y: Discrete Response Models for Panel Data
5.1.1 Ignoring the Panel Structure
5.1.1.1 The Pooled Linear Probability Model
5.1.1.2 Pooled Logistic Regression
5.1.1.3 Pooled Probit Regression
5.1.2 Modeling the Panel Structure
5.1.2.1 Correlated Heterogeneity: Fixed Effects Estimation
5.1.2.2 Uncorrelated Heterogeneity: Random Effects Estimation
5.1.2.3 Choosing Between Pooled, Fixed, and Random Effects Estimation
When Do We Apply Pooled, FE, and RE Estimation?
A Hybrid Model
5.1.3 Extensions
5.2 Modeling the Change of Y: Discrete-Time Event History Models for Panel Data
5.2.1 Basic Terminology
5.2.2 How to Estimate a Discrete-Time Hazard Model
5.2.2.1 OLS Regression
5.2.2.2 Logistic Regression
5.2.2.3 Logistic Discrete-Time Hazard Model
5.2.3 Applying the Discrete-Time Event History Model
5.2.3.1 Non-repeatable Singular Events
5.2.3.2 Unobserved Heterogeneity in Event History Models
5.2.3.3 Uncorrelated Heterogeneity: Random Effects Event History Models
5.2.3.4 Correlated Heterogeneity: Fixed Effects Event History Models
5.2.3.5 Applying Continuous-Time Event History Models Within a Panel Design
5.2.4 Extensions
5.3 Conclusion and Further Reading
Chapter 6: How to Do Your Own Panel Analysis
Chapter 7: Useful Background Information
7.1 Functions of Random Variables
7.2 Estimation and Testing
7.2.1 Ordinary Least Squares
7.2.1.1 How to Compute a Regression Model Fitting the Data?
7.2.1.2 Sampling and Sampling Errors
7.2.1.3 How to Choose Between Different Estimation Methods?
7.2.1.4 How to Estimate the Parameters of an Unknown Population with a Sample of Data?
7.2.1.5 How to Test Parameters of an Unknown Population with a Sample of Data?
7.2.2 Maximum Likelihood
7.3 Web Site of the Textbook
References
Index
Author Index